大模型对齐与安全

Measuring Corporate Human Capital Disclosures: Lexicon, Data, Code, and Research Opportunities

Elizabeth Demers, Victor Xiaoqi Wang, Kean Wu

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 50 pages, 6 figures, 5 tables

Journal ref Journal of Information Systems 38 (2024) 163-186

Human-like object concept representations emerge naturally in multimodal large language models

Changde Du, Kaicheng Fu, Bincheng Wen, Yi Sun, Jie Peng, Wei Wei, Ying Gao, Shengpei Wang, Chuncheng Zhang, Jinpeng Li, Shuang Qiu, Le Chang, Huiguang He

机构 * State Key Laboratory of Brain Cognition and Brain-inspired Intelligence Technology（脑认知与脑启发智能技术重点实验室） ； Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所） ； School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院） ； Institute of Neuroscience, State Key Laboratory of Brain Cognition and Brain-Inspired Intelligence Technology（神经科学研究所） ； CAS Center for Excellence in Brain Science and Intelligence Technology（中国科学院脑科学与智能技术卓越创新中心） ； Chinese Academy of Sciences（中国科学院） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Published on Nature Machine Intelligence

Journal ref Nature Machine Intelligence, 2025

Multilingual Hate Speech Detection in Social Media Using Translation-Based Approaches with Large Language Models

Muhammad Usman, Muhammad Ahmad, M. Shahiki Tash, Irina Gelbukh, Rolando Quintero Tellez, Grigori Sidorov

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

Linear Representation Transferability Hypothesis: Leveraging Small Models to Steer Large Models

Femi Bello, Anubrata Das, Fanzhi Zeng, Fangcong Yin, Liu Leqi

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Symmetrical Visual Contrastive Optimization: Aligning Vision-Language Models with Minimal Contrastive Images

Shengguang Wu, Fan-Yun Sun, Kaiyue Wen, Nick Haber

机构 * Stanford University（斯坦福大学） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Accepted to ACL 2025 Main. Project Website: https://s-vco.github.io/

Explainable AI Systems Must Be Contestable: Here's How to Make It Happen

Catarina Moreira, Anna Palatkina, Dacia Braca, Dylan M. Walsh, Peter J. Leihn, Fang Chen, Nina C. Hubig

机构 * Data Science Institute UTS Australia（UTS澳大利亚数据科学研究院） ； IT:U Austria（奥地利IT:U学院） 纠错

专题命中 其他安全 ：safety(abstract)；分类 cs.AI、cs.CY、cs.LG

KnowCoder-X: Boosting Multilingual Information Extraction via Code

Yuxin Zuo, Wenxuan Jiang, Wenxuan Liu, Zixuan Li, Long Bai, Hanbin Wang, Yutao Zeng, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng

机构 * Key Laboratory of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences（网络数据科学与技术重点实验室，计算技术研究所，中国科学院） ； State Key Laboratory of AI Safety（人工智能安全国家重点实验室） ； School of Computer Science, University of Chinese Academy of Sciences（中国科学院大学计算机科学学院） ； School of Software, Northeastern University（东北大学软件学院） ； School of Software, Peking University（北京大学软件学院） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments ACL 2025 Findings

The Price of Format: Diversity Collapse in LLMs

Longfei Yun, Chenyang An, Zilong Wang, Letian Peng, Jingbo Shang

机构 * University of California, San Diego（加州大学圣迭戈分校） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 14 pages, 7 figures

LoRASuite: Efficient LoRA Adaptation Across Large Language Model Upgrades

Yanan Li, Fanxu Meng, Muhan Zhang, Shiai Zhu, Shangguang Wang, Mengwei Xu

机构 * Beijing University of Posts and Telecommunications（北京邮电大学） ； Peking University（北京大学） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Structured Prompting and Feedback-Guided Reasoning with LLMs for Data Interpretation

Amit Rath

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 21 pages, 2 figures

Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling

Kaiwen Zheng, Yongxin Chen, Hanzi Mao, Ming-Yu Liu, Jun Zhu, Qinsheng Zhang

机构 * Department of Computer Science & Technology, Institute for AI, Tsinghua University（1 计算机科学与技术系，人工智能研究院，清华大学） ； NVIDIA（2 英特尔） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Accepted at ICLR 2025

Certifying Knowledge Comprehension in LLMs

Isha Chaudhary, Vedaant V. Jain, Gagandeep Singh

机构 * Siebel School of Computing and Data Science University of Illinois, Urbana-Champaign（计算机与数据科学学院 伊利诺伊大学厄巴纳-香槟分校） 纠错

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

LanguageMPC: Large Language Models as Decision Makers for Autonomous Driving

Hao Sha, Yao Mu, Yuxuan Jiang, Li Chen, Chenfeng Xu, Ping Luo, Shengbo Eben Li, Masayoshi Tomizuka, Wei Zhan, Mingyu Ding

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

SAEs $\textit{Can}$ Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs

Aashiq Muhamed, Jacopo Bonato, Mona Diab, Virginia Smith

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning

Nikhil Shivakumar Nayak, Krishnateja Killamsetty, Ligong Han, Abhishek Bhandwaldar, Prateek Chanda, Kai Xu, Hao Wang, Aldo Pareja, Oleg Silkin, Mustafa Eyceoz, Akash Srivastava

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 25 pages, 13 figures, 6 tables

TULIP: Towards Unified Language-Image Pretraining

Zineng Tang, Long Lian, Seun Eisape, XuDong Wang, Roei Herzig, Adam Yala, Alane Suhr, Trevor Darrell, David M. Chan

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments (v2) Clarified fine-tuning process, updated appendix

Triple Phase Transitions: Understanding the Learning Dynamics of Large Language Models from a Neuroscience Perspective

Yuko Nakagi, Keigo Tada, Sota Yoshino, Shinji Nishimoto, Yu Takagi

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 46 pages

Analyzing and Boosting the Power of Fine-Grained Visual Recognition for Multi-modal Large Language Models

Hulingxiao He, Geng Li, Zijun Geng, Jinglin Xu, Yuxin Peng

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Published as a conference paper at ICLR 2025. The model is available at https://huggingface.co/StevenHH2000/Finedefics

RAP: Retrieval-Augmented Personalization for Multimodal Large Language Models

Haoran Hao, Jiaming Han, Changsheng Li, Yu-Feng Li, Xiangyu Yue

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Accepted by CVPR 2025. Code: https://github.com/Hoar012/RAP-MLLM

Training Domain Draft Models for Speculative Decoding: Best Practices and Insights

Fenglu Hong, Ravi Raju, Jonathan Lingjie Li, Bo Li, Urmish Thakker, Avinash Ravichandran, Swayambhoo Jain, Changran Hu

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Published as a workshop paper at SCOPE - ICLR 2025

Towards Aligning Language Models with Textual Feedback

Saüc Abadal Lloret, Shehzaad Dhuliawala, Keerthiram Murugesan, Mrinmaya Sachan

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Accepted to EMNLP 2024

3D-MolT5: Leveraging Discrete Structural Information for Molecule-Text Modeling

Qizhi Pei, Rui Yan, Kaiyuan Gao, Jinhua Zhu, Lijun Wu

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Accepted by ICLR 2025

Cream of the Crop: Harvesting Rich, Scalable and Transferable Multi-Modal Data for Instruction Fine-Tuning

Mengyao Lyu, Yan Li, Huasong Zhong, Wenhao Yang, Hui Chen, Jungong Han, Guiguang Ding, Zhenheng Yang

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments update comparison with sota and analysis

L3Ms -- Lagrange Large Language Models

Guneet S. Dhillon, Xingjian Shi, Yee Whye Teh, Alex Smola

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments International Conference on Learning Representations (ICLR), 2025

Is a Good Foundation Necessary for Efficient Reinforcement Learning? The Computational Role of the Base Model in Exploration

Dylan J. Foster, Zakaria Mhammedi, Dhruv Rohatgi

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments V2: Improved number of prompts used by Algorithm 1

Utilizing Large Language Models to Synthesize Product Desirability Datasets

John D. Hastings, Sherri Weitl-Harms, Joseph Doty, Zachary J. Myers, Warren Thompson

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 9 pages, 2 figures, 6 tables, updated author list

Journal ref 2024 IEEE International Conference on Big Data (IEEE BigData 2024)

JAM: Controllable and Responsible Text Generation via Causal Reasoning and Latent Vector Manipulation

Yingbing Huang, Deming Chen, Abhishek K. Umrawal

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 10 pages, 3 figures, and 6 tables

Non-Halting Queries: Exploiting Fixed Points in LLMs

Ghaith Hammouri, Kemal Derya, Berk Sunar

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG

Training AI to be Loyal

Sewoong Oh, Himanshu Tyagi, Pramod Viswanath

专题命中 其他安全 ：alignment(abstract)；分类 cs.AI、cs.CY、cs.LG

Comments 13 pages 4 figures

Flaming-hot Initiation with Regular Execution Sampling for Large Language Models

Weizhe Chen, Zhicheng Zhang, Guanlin Liu, Renjie Zheng, Wenlei Shi, Chen Dun, Zheng Wu, Xing Jin, Lin Yan

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI、cs.LG